Story perspectives
Amodei Proposes Third-Party AI Safety Evaluators Inside Firms
By Drooid · · How we work
1 of 3
Amodei Proposes Evaluators
- Anthropic CEO Dario Amodei proposed embedding third-party safety evaluators inside frontier AI firms.
- He outlined evaluator desks, badges, and laptops in a September 12, 2026 essay.
- OpenAI CEO Sam Altman reposted the plan and promised similar evaluator access at OpenAI.
- President Donald Trump rebuked Amodei on Truth Social on September 14, 2026, dismissing extra AI guardrails.
- Critics Julie Andersen Hill and Deborah Raji argue evaluators lack enforcement power and may face conflicts of interest.
1 / 3
